On Sun, 22 Nov 2015 22:38:50 -0600, Ed Gould wrote: >http://www.theregister.co.uk/2015/11/22/ >world_needs_252288000_seconds_to_decide_fate_of_leap_seconds/ > I'd oppose any chahge to existing conventions. There are 3 desiderata of a time standard:
o Continuity o Atomic precision o Close aggreement with earth rotation. Only any two of those can be achieved concurrently. We have 3 systems respectively supporting two at a time: o UTC: precise and earth-compatible; lacking continuity o UT1: continuous and earth-compatible; lacking precision o TAI: precise and continuous; lacking earth-compatibility. No new convention need be introduced. POSIX made an adverse selection of UTC: leap seconds are a problem. For practical purposes UT1 would have been better. But POSIX's unwise choice is not rationale for changing the specification of UTC. If necessary, POSIX should be changed to specify either TAI or UT1. MVS also chose UTC and accommodated leap seconds with CVTLSO. But IBM failed to follow through in that design: IBM failed to enhance time conversion facilities such as STCKCONV and CONVTOD to make them leap-second savvy. It's a half-baked recipe. -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
